View
228
Download
0
Category
Preview:
Citation preview
A comparison of the determinants of childbearing intentions before and after
the crisis in Europe
Lydia Palumbo
European Doctoral School of Demography
and
Emiliano Sironi
Università Cattolica del Sacro Cuore
Abstract
Using the data from the second and from the fifth rounds of the European Social Survey that took place
respectively before (2004) and after the beginning of the Great Recession (2010), we aim at
investigating whether the crisis has affected the intentions of having a child within three years from the
interview.
Hence, we have implemented a multilevel regression model, combining micro level and macro level
predictors of childbearing. We do not find evidence of a generalized decline in the intentions of having a
child, but we obtained that the effect depends on the employment status of individuals. In more detail,
the effect of employed significantly interacts with Global Crisis; individuals that are currently
unemployed seem to be less likely to have a child within the next three years after the beginning of the
crisis. Results are uncertain for temporary employees that seem to be more intentioned to have children
than unemployed ones, but less likely to be intentioned than permanent employees during the economic
crisis.
Keywords: Intentions, Childbearing, European Social Survey, Great Recession JEL: J11, J13
1. Introduction
The transition to adulthood is a complex process involving several steps that include: bringing school
education to completion, entering into the labour market, leaving the parental home, being in a stable
relationship and parenthood (Hogan and Astone 1986, Goldscheider and Goldscheider, 1993).
One of the most important changes in western countries in the last decades is the continuous
postponement of family formation: the fertility level has dramatically decreased in all developed
countries (Adserà, 2004).
If economic constraints and difficulties in finding a stable job tend to have a negative impact on the
transition process to adulthood and on family formation, the recent Global Crisis may be expected to
further worsen the conditions of European young people, engendering an additional delay in their life
course path. There are some evidences that countries most affected by the last great recession have
experienced a decrease in fertility too, compared with a continuation of recent trends, especially at
younger ages (Goldstein et al, 2013). Moreover, the employment instability, one of the most serious
consequences of economic crisis, has been proved to have a negative impact on final number of children
for both men and women (Ciganda, 2015).
In this study we compare fertility intentions before and after the beginning of the “Great Recession” in
order to study decision-making process of having a child.
Comparing data from the second (2004) and the fifth (2010) rounds of the European Social Survey
(ESS) we try to assess whether the Global crisis has affected the fertility intentions. In addition we test
also whether some key determinants of fertility intentions modified their impact during the crisis.
Because of the interaction between individual determinants and macroeconomic context we develop a
multilevel approach that takes into account both the effect of micro-level and macro-level determinants
After that short introduction the reminder of the paper is organized as follows. Section two presents the
main literature review investigating the relationship between economic crisis and fertility intentions.
Section 3 presents the methodology of the study that includes the sample design, the estimation strategy
and the variables used in the models. Section 4 discusses the empirical results, while section 5
concludes.
2. Literature Review
The economic recession is a sudden period of decline in economic activity that lasts longer than a few
months (Sobotka, 2011). This period could easily lead to a temporary decrease in fertility due to a bad
economic situation characterized by aspects like lower income and increased job demands. One possible
cause of this decline is temporary postponement, which is usually offset during times of economic
prosperity (Sobotka, 2011).
Scholars have often dealt with the association between fertility and economic growth (and, thus,
recession) at both theoretical and empirical level.
On the theoretical side, the micro-founded neoclassical economic theory developed by Becker (1960)
analyzes this relationship by considering children as durable goods. This means that they provide utility
and imply opportunity costs. This model points out that childbearing involves two opposite effects, a
substitution and an income one: the first one implies that the demand for children diminishes as soon as
their opportunity cost, their price, increases and the latter one means that the demand of children
increases as the household income increases.
According to Easterlin (1980, 1995), parents affect their sons’ and daughters’ fertility through the
relative income, e.g. income potential over the expected standard of living. This means that an economic
recession could reduce this index and cause a delay of marriage and childbearing.
On the empirical side, the most studied relationship in a period of recession regards the one between
women’s participation to labor market and fertility and there is evidence of both a positive and negative
association.
Butz and Ward (1979) showed the counter-cyclicality of aggregate economic activity in the 1970s,
especially women’s rising employment, on fertility rates in the United States. According to them, the
substitution effect hypothesized by Becker would have prevailed over the income one. Thus, women
would have chosen work over childbirth.
More recently, Macunovich (1995) reaffirmed the American pro-cyclicality, since the negative effect of
a decrease of unemployment during a period of economic crisis on fertility overcame the positive effects
of the lower price of women’s time during a recession period; several studies support Macunovich’s
relationship, not only in the American context (Ahn, 2001 in Spain; Kravdal, 1994 in Norway; Hoem et
al., 2000 in Sweden).
Sobotka et al. (2011) tried to conciliate these positions stating that different results may mean that there
are subjects who find convenient having a child during crisis times and some others who think that it is
good to delay the next birth or avoid having a child.
The determinants of fertility related to macroeconomic conditions are usually GDP, either its level or its
percentage growth, and unemployment. GDP is an indicator of the overall economic well-being and
welfare in average terms, although it is not a perfect substitute (Martin, 2002). Adsera (2004)
determined a negative relationship between GDP per capita and fertility using a panel of OECD
countries. Through the analysis of the negative relationship between Australian GDP and
unemployment and Australian TFRs, Martin (2002) argued that the magnitude of this negative
relationship depended positively on macroeconomic conditions: when there were years of economic
recession, there was a pronounced decline of fertility, while, when there was strong economic growth,
the decrease of TFRs was much slower. Santow et al. (2001) also found out that in Sweden, where there
was a fall in GDP per capita and a rise in unemployment in 1990s, GDP per capita was highly
significant and that economic recession might have affected negatively first birth conception. According
to Billingsley (2010), GDP decline in Post-Soviet countries depressed fertility and improved economic
performance increased the likelihood of postponement. Adsera (2011), when studying the effects of
economic uncertainty on fertility in South America, found out that the economic downturns tended to
slowdown first birth conceptions. However, they also were able to understand that GDP growth affected
significantly (and positively) first birth rates only when unemployment was not part of the implemented
model, meaning that the actual influential determinant was redundancy.
Luci et al. (2010) found out a J-shaped relationship between GDP per capita (at constant prices) and
fertility in developed countries. This would mean that after a certain threshold of wealth (represented by
GDP per capita), the relationship between GDP growth and fertility is positive. However, this
association does not hold in all the advanced economies, since it depends on the specific situation of the
country, especially the institutional context.
Unemployment is the most used variable when discussing the effects of economic conditions on
fertility. As Sobotka (2011, p.11) notices: “[…] differently from the GDP change, unemployment
growth constitutes a more tangible indicator of the impact of economic crisis which has a direct bearing
on women and men of reproductive age […]”.
Orsal et al. (2010) suggested that unemployment rate was a better indicator of economic influence on
fertility, because an individual gives more importance to the possibility of being employed or rather than
to the growth in the economy when deciding whether to have a child or not. They reached the
conclusion that there is a positive association between unemployment and fertility in OECD countries
from 1976 to 2008, even when there was a distinction by gender. Female unemployment used to
increase its importance over time, while male did not.
Sobotka et al. (2011) argued that being unemployed among childless men affected negatively the
propensity to become a father, since the breadwinning role of the man is still very important to provide
for the whole family. They attribute the resistance of this traditional role to the fact that women that
usually take a pause when they have a child and parental allowances are not usually enough to offset the
lost source of income. This theoretical view has also been confirmed by empirical studies (Ahn et al.
(2001) in Spain). This association is indeed much more ambiguous for women, since losing a job makes
childbearing opportunity costs lower, thus, more convenient. Thus, economic recession could also be
associated with higher fertility.
However, many studies show the prevalence of an income and uncertainty effect, which discourages
fertility when an unemployment condition comes up. Adsera (2004) found out a negative relationship
between unemployment and fertility rates in a sample of OECD countries. Adsera (2004) pointed out
that the proportion of unemployed people and the average length of unemployment condition affect
fertility. This relationship is moderated by institutional characteristics of a country, as well as by the
characteristics of the labour market: in particular, its flexibility and its ability to generate employment
are elements that encourage fertility. Thus, countries with more female labour force participation in
Europe were able to overcome lower fertility rates due to higher presence of women at work through the
higher flexibility of the market (United Kingdom and Netherlands) or larger public sector (France and
Nordic countries). On the opposite, countries with low female participation, small government sectors
and rigid markets have not found the way to reduce uncertainty related to unemployment. Schimtt
(2008) also remarked that a woman’s unemployment condition and the presence of a rigid labour
market, leading to a long – term unemployment, might be an opportune condition leading to a decision
in favour of births, since the duration of a woman’s jobless situation overcomes her human capital
accumulated through the years. The presence of generous unemployment benefits, as well as cash
pecuniary benefits, is usually considered favourable to fertility decisions. The evidence of it is
particularly present in Nordic countries (Hoem, 2000, Andersson, 2000 in Sweden; Aassve et al, 2009 in
Norway; Vikat, 2004 in Finland).
Besides unemployment, an important consequence of a recession is employment instability, which leads
to more layoffs, more part-time contracts and time-limited work contracts. Ranjan (1999) developed a
model linking income uncertainty and fertility and the future level of income of the individual (which is
related to uncertainty). She pointed out that those who are below certain predicted thresholds of income
are more likely to delay childbirth when the future is uncertain, since they wait for the resolution of the
uncertainty. This is also what happened in Post-Soviet countries after 1989, where an increase in
uncertainty and a consequent rise in the income threshold, pushed more and more people below the
threshold. Bernardi et al. (2008) studied the impact of uncertainty on fertility in Western and Eastern
Germany after the reunification in 1989, since Western Germany was affected by uncertainty because of
the increase of temporary contracts and job mobility, whereas Eastern Germany went through
unprecedented levels of unemployment after the reunification. This modified the balance of family
formation and family enlargement because it required working harder and longer to obtain a permanent
contract with the result of giving up the beginning of a long-term activity, such as childbearing. In
Eastern Germany, Kreyenfeld (2010) showed that economic uncertainty might have had different effects
according to the level of education: unemployed women with higher education tend to delay their
motherhood (although this is a weak relationship), while unemployed women with lower education tend
to use this period become mothers and, then reduce the uncertainty of their life.
The presence of an economic transition in Hungary and Bulgaria led also to uncertainty (anomie), which
made people indecisive and persuade them to postpone their childbirth (Billari, 2006). There is also
evidence that those who are usually more affected by economic uncertainty are young working age men
and women, who are going through a deterioration of the relative economic conditions (Sobotka, 2011).
The use of more precarious forms of contracts, such as temporary ones, which is mostly present in
Southern Europe, can also be one of the causes of the mismatch between desired and actual fertility
(Adsera, 2004; Dolado et al. in Spain, 2002; Modena et al., 2011).
The relationship between the current crisis and fertility decisions has not been widely studied, since the
argument is quite recent. Goldstein et al. (2013) after the beginning of the recent economic recession
analyzed the effect of the economic crisis begun in 2008 on fertility at a macroeconomic level. They
affirmed that the recession in 2008 took place when there was a modest increase in period fertility rates
of European countries. However, the response of different states to the deterioration of economic
conditions was not unique. Goldstein et al. (2013) found out a negative impact of unemployment on
fertility, which affected younger generations and their first births together with older generations and
second births. They explained the first result arguing that fertility plans may be better revised when
people are young than when they are constrained by biological limits of fertility. The second result
might be due to the fact that younger people who have a second birth when they are young are also less
educated and less career-oriented, whereas people who are going to have a second or third child at older
ages are usually highly educated and take into account a downturn in economic recession. As well as a
timing effect in the decline of births, Goldstein et al. (2009) detected also a quantum effect on fertility:
other factors might have had an impact on TFRs, such as institutional factors (parental leaves policies
and welfare support) or the exploitation of the advantage of temporary non-employment. These factors
allowed economic crisis to hit countries in a different way. For example, on the one hand, the fertility of
Baltic and Central states was quite untouched by the crisis, because institutions may have softened
adverse economic conditions. On the other hand, Southern states were particularly affected by limited
family policies and inability to protect younger people from economic uncertainty and this might have
worsened fertility conditions. Testa and Basten (2012) studied the impact of socio – economic
conditions of citizens in 27 European countries on lifetime fertility intentions. They performed a
multilevel analysis joining macro and micro level measures. They were able to detect the negative
impact of the economic recession on fertility intentions due to a perceived worsening of economic
conditions and to an increase of youth unemployment. This was especially true in countries with bad
economic conditions, which led to a sovereign debt crisis, such as the “PIIGS”. This situation required
the introduction of austerity packages, which often had an impact on social spending and family policies
and, thus, increase the perceived uncertainty. However, the results of a decline in fertility intentions are
visible in Greece, but there is not enough data to state the presence of an effect due to recession in other
regions.
3. Methodology
The sample considered for the empirical analysis consists of the data collected by the European Social
Survey (ESS) in 2004 and 2010. These rounds were respectively the second and the fifth.
The ESS submits a questionnaire to citizens of different European countries every two years. It includes
both core questions on socio-demographic information of the individuals and a rotating module with
different questions on different topics for each round.
The choice of the second and the fifth round rests on the fact that they are the only ones that comprise a
question on fertility intentions and that were submitted before and after the beginning of the economic
crisis in 2008.
The joined sample (which includes both data from 2004 and 2010) consists of more than 80,000
individuals belonging to 21 European countries and interviewed in the two rounds. They are Belgium,
Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Netherlands,
Norway, Poland, Portugal, Slovenia, Slovakia, Spain, Sweden, Switzerland, United Kingdom and
Ukraine. An important remark is that data is not longitudinal. Therefore, the individuals interviewed in
2004 are not the same ones interviewed in 2010.
In our paper we decided to restrict the analysis to those individuals who were affected by decisions
regarding fertility intentions. Therefore, we chose only subjects whose age was between 18 and 40 years
old at the time of the interview were part of the final sample; observations are slightly less than 26,000
and they are 13,460 in 2004 and 12489 in 2010. There is a mild overrepresentation of 2004 over 2010
due to the reduction of the sample (see tab. 1), but there is no change in the proportion of the
representation of the countries. Therefore, the sample is very well balanced.
Table 1. The composition of the sample by year
Year of the analysis Frequency Percentage 2004 13,460 51.87
2010 12,489 48.13
25,949 100
3.1. Variables
Dependent variable. The dependent variable regards short-term fertility intentions. It contains the
answers to the question “Do you plan to have a child within the next three years?” and allows four
possible ordered answers: “Definitely not”, “Probably not”, “Probably yes” and “Definitely yes”. Then,
these variables represent timing fertility intentions, which, according to the theory, are the best predictor
for actual fertility (Miller and Pasta, 1995). This variable presents 2230 missing values. This may have
happened due to the difficulty of answering to this kind of questions, since it is quite hard to make plans
in the short-term. Missing observations were indeed dropped.
Independent variables. These variables can be divided into two big groups: macro and micro-level
variables. The first ones include country-level indicators, as the unemployment rate and the percentage
change of GDP, and the other ones comprise individual-level characteristics (the ones collected by the
ESS survey).
Macro level variables
Unemployment rate. The source of unemployment rate is Eurostat1. The percentages date at 2004 and
2010. According to Testa and Basten (2012), it is important to include unemployment because this
measure gives a tangible indicator of the impact of crisis on fertility. In fact, the low fertility in Southern
Europe could also be attributed to low levels of employment and job instability. For example, Adsera
(2004) found out a negative effect of female unemployment, male unemployment and both female and
male unemployment on TFR using a panel of 23 OECD countries. She also found a negative effect of
youth (male and female) unemployment.
Therefore, the expected sign of the level of unemployment is a negative one.
GDP percentage annual growth. This variable indicates the percentage annual growth of GDP at
purchasing power parity with constant prices (the reference year is 2005) for both 2004 and 2010 (with
respect to 2003 and 2009). The source of data is OECD2. Unfortunately, it is not possible to control for
the level of GDP due to its higher correlation with unemployment rate. The expected sign of the relation
between fertility intentions and GDP change can be hypothesized through the analyses of Billari et al.
(2009) and Luci et al. (2010), whose results show a positive relationship in developed countries. Since
the majority part of the nations in the sample are considered advanced economies, the expected sign of
this variable is thought to be positive.
Testa et al. (2012) consider GDP per capita as the reflection of the cross-country differences in
socioeconomic conditions at the time when the respondents reported their fertility intentions (in this
case, there would be the analysis of the change of the socio-economic background). Moreover, they
found a positive effect of GDP on fertility. Santow et al. (2001) also used GDP change as a country-
level variable in a study in Sweden and it resulted to have a positive relation with childbirth rates.
Micro level variables
Age. This variable is divided into 4 categories. The first one comprises people from 18 to 25 years old,
the second people from 25 to 30, the third goes from 30 years old to 35 and the fourth from 35 to 40.
The reference category is the youngest one. Because of childbirth delay and biological constraints, the
sign associated with the coefficient of age is expected to be positive in the older categories (Perelli
Harris, 2006).
Gender. This is an important variable when talking about fertility intentions. This variable is represented
by a dummy that takes value 0 if the individual is a man and 1 if not. There are different preferences
among women and men, especially in the last decades when women were more educated and had more
career prospects (Van De Kaa, 2008).
Presence of other children. Presence of children is measured through a dummy that takes value 1 if the
individual has children and 2 if not.
1 http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=une_rt_a&lang=en 2 http://stats.oecd.org/index.aspx?queryid=60707
Education of the individual. Education is classified according to ISCED-97 classification. This variable
is a four-level categorical variable. The lower category comprises individuals with less lower secondary
education (ISCED 0-1). The higher one involves individuals with lower secondary education completed
(ISCED 2). The higher one joined completed upper secondary education to completed post-secondary
non-tertiary education (ISCED 3-4). The highest and fourth category concerns completed tertiary
education (ISCED 5). As scholars notice, the effect of education on fertility intentions could be
ambiguous. On one hand, Axinn and Barber (2001) argue that education should have a negative impact
on fertility, especially women’s, because it allows them to have new opportunities, which could replace
or at least lower the role of motherhood. However, they found little evidence supporting this thesis. On
the other hand, Perelli-Harris (2006) argues that education could lower uncertainty because a person
who is more educated was able to acquire the knowledge to solve problems. Thus, these people should
be more secure and have higher fertility intentions.
Moreover, the historical context is also significant. Many scholars who studied the patterns of second
demographic transition argued that a decline in fertility and therefore fertility plans in the nineties was
due to higher women’s education. Education could have led to higher career prospective and, therefore,
to delaying or forgoing childbirth (Van De Kaa, 2008), to a shift in family values or to a better
knowledge of contraception (Axinn et al., 1994). However, the relationship among women’s education
and fertility plans seemed to have turned positive over the years since the substitution effect that women
had to deal with, when having high education and, therefore, higher career perspectives and ambitions,
(Becker et al., 1960) was overcome by an income effect (Testa, 2012). Furthermore, in 2004 and 2010,
there was a recoupment of fertility among those couples of who delayed it (Goldstein et al., 2009).
People that are more educated represented a big part of them (Mclanahan, 2004). Therefore, their
fertility intentions in 2004 and 2010 could have increased. The expected sign of this variable effect is,
therefore, uncertain.
Marital status. This variable comprises three categories: being married or living in a civil union
partnership, not being married and a residual category, which considers widows, divorced and separated
people, i.e. not married anymore. This variable was quite difficult to define because the methodology
that the ESS used changed between 2004 and 2010. From 2006 on, the ESS questionnaire has included
people living in a civil union beyond the categories used before (married, separated, divorced, widowed
and never married). Unfortunately, there was not a precise a criterion of classification of this category
before 2006. Therefore, it was not possible to understand whether they were considered married or
never married. In this paper, the variable has been recoded in the most natural way: people living in a
civil union were added to married people and, thus, they constitute the category “Married or in a civil
union partnership” along with them.
Moreover, it was not possible to find out a distinction between people who cohabitate and single. Prioux
(in the context of France, 2003) and Billari et al. (2007) consider marital status a key variable to reach
home leaving and autonomy, even though it is not fundamental to have a stable relationship to leave
parental home. Moreover, Gamundi et al. (2009) argue that cohabiting people have different preferences
from the married ones. For instance, they are less likely to have children. Therefore, married couples or
those living in a civil union are expected to have higher fertility intentions than unmarried ones.
Country of origin. This variable is a dichotomous dummy, which considers whether an individual was
born or not in a certain country. It is important to notice that this variable gives the only effect of being a
first-generation immigrant. According to the theory, there are different hypotheses about the effect of
immigration on fertility intentions, which predict opposite influences (Kulu, 2005). Immigration of first
generation depends generally on the country of origin and on the type of immigration (getting higher
education, high profile jobs or achieving better life conditions) (Rosenwaike, 1973; Stephen and Bean,
1992; Kahn, 1994). Therefore, the expected sign is uncertain.
Belonging to an ethnic minority. This variable is dichotomized into two categories: belonging to an
ethnic minority and not belonging. Goldscheider et al. (1971) predicts two different kinds of behavior
related to ethnicity. On one the hand, being part of a minority could hinder fertility intentions in order to
lower barriers between the majority part of the population and the minority itself. On the other hand,
minorities could prefer avoiding fertility controls and choose larger families as an instrument to separate
them from the others. Therefore, the sign of the coefficient is uncertain.
Occupational status. This variable is categorical and consists of six groups. The first one is represented
by employees with a permanent contract, the second one by those with a temporary (or fixed term)
contract and the third one by self-employees. The fourth category includes students, the fifth
unemployed (with no distinction between those looking for a job and not looking for a job) and the sixth
one is a residual category comprising those dealing with housework, community or military service and
retired. When computing the model, the reference category will be the one of employees with regular
contract, i.e. the most secure form of occupational status. The coefficient related to the category of
temporary employees and students is expected to be negative (with respect to the coefficient of the
regular employees), the one of self-employees is thought to be uncertain. On a macro level, Adsera
(2004) found a negative effect of the share of self- employment on fertility rates in a sample of OECD
countries. She motivated the sign saying that self-employees, with the exception of high-level
professionals, have indeed more flexible hours, but they are at the bottom of earnings distribution, they
face more income uncertainty and are less able to take advantage of parental leaves. However, Modena
et al. (2013) find a positive effect of being self-employed on child-timing fertility intentions in Italy
because they are mainly professionals and entrepreneurs. The presence of an ambivalent effect could
also hold for unemployed people. Unemployment is usually a condition that discourages fertility or
fertility intentions with respect to employed people, at macro and micro level (Adsera, 2004; Testa and
Basten, 2012), but there could be the possibility that some people exploit this period of employment
uncertainty to have a(nother) child (Schmitt, 2008).
Well being (happiness). This is a continuous scale variable. People were asked how happy they were.
They had the chance of pointing out the level of their happiness through a Likert-scale, going from 0 to
10. According to Perelli-Harris (2006), in Russia, well-being has a positive influence on men and
women’s actual fertility and women‘s well-being has a significant influence also on desired fertility.
With regard to men’s happiness, it resulted to have a positive influence on women’s fertility desires, but
there is a lack of information on its effect on men’s fertility preferences.
Gender role. Gender role is an index derived through the implementation of the principal component
analysis with two items of the questionnaire on gender roles in the society. The first one is “A woman
should be prepared to cut down on paid work for sake of family” and the second one is “When jobs are
scarce, men should have more right to a job than women.” These items gave the possibility of answering
to these questions through a Likert scale, from one, which means strong agreement, to ten, which means
total disagreement. This means that the lower the score assigned to an individual, the higher the
traditional idea of gender roles. The answers “Don’t know” were recoded as a middle point endorsement
instead of missing values. This factor is quite important, since several scholars that study the second
demographic transition affirmed that the change of the woman’s role was an influential factor in
determining the birth decline in the nineties (Van De Kaa, 2002; Lesthaege, 2010; Maclahnan, 2004).
Religiosity. Mcquillan (2004) focuses on the relationship between fertility and religion. He thinks that
three main aspects are important to analyze religion: the first one concerns religion and values, then the
second one religious institutions and the third one religious identity. Values are important, although
their relationship with religion is overrepresented, and it is important to specify in which way religious
beliefs influence fertility levels.
3.2. The model
The model used to analyze the relationship between the above-mentioned variables and child-timing
fertility intentions is a linear random-intercept ordered logistic regression, since the dependant variable
is categorical and ordinal. The use of this method comes for the reason that the units of observations in
the sample are naturally clustered, as the individuals belong to different countries. Hence, their
observations might be related or dependant. We decided to compute two models including only
individual level variables, the first one without the variables indicating occupational status and well-
being and the second one comprising them. Beyond them, we decided to introduce two macroeconomic
variables at country level, i.e. the annual percentage change of GDP per capita and the level of
unemployment rate of the considered year. In fact, as Testa (2012) argues, models on more than one
level are used “to represent the complex causal process underlying the behavior of individuals in a
social context and to draw valid inferences regarding the relationships at the relevant hierarchical
levels”. Thus, considering the specific situation of our paper, the rising unemployment rates and the
potential reduction of the annual GDP growth in the considered countries might have had an effect both
on fertility intentions and on the analyzed subjective characteristics, which also have an effect on the
variable of interest (and creating a filtered effect). Thus, we are interested in understanding, once the
differences among countries are taken into consideration, whether the worsening of macroeconomic
conditions or personal socio-economic situation, as well as cultural factors, affect fertility intentions.
Finally, the computed models are four:
(1) Fertility intentionsij=Marital statusij+ Gender role indexij + Being religiousij + Personal Educationij+
Ethnic Minorityij+ Born in the countryij+ Year of the interviewij+ Being childlessij+ Genderij+ Ageij+ εij
(2) Fertility intentionsij= Marital statusij+ Gender role indexij + Being religiousij + Personal Educationij+
Ethnic Minorityij+ Born in the countryij+ Year of the interviewij+ Occupational statusij+ Happinessij+
Occupational statusij* Year of the interviewij+ Happinessij*year of the interviewij+ Being childlessij+
Genderij+ Ageij+ εij
(3) Fertility intentionsij= Unemployment ratej + Percentage change of GDPj+ Marital statusij+ Gender role
indexij + Being religiousij + Personal Educationij+ Ethnic Minorityij+ Born in the countryij+ Year of the
interviewij + Being childlessij+ Genderij+ Ageij+ εij
(4) Fertility intentionsij= Unemployment ratej + Percentage change of GDPj+ Marital statusij+ Gender role
indexij + Being religiousij + Personal Educationij+ Ethnic Minorityij+ Born in the countryij+ Year of the
interviewij+ Occupational statusij+ Happinessij+ Occupational statusij* Year of the interviewij+
Happinessij*year of the interviewij+ Being childlessij+ Genderij+ Ageij+ εij
4. Empirical Results
In table 2, we display the results of two models that have been computed at individual level, without
considering the macroeconomic variables that had been mentioned in the following subsection. The first
model is a basic one and does not contain the potential endogenous variables, i.e. the occupational status
and the well-being ones, together with their interactions, while the second one includes those
determinants.
Table 2. Models with only individual characteristics – Intentions of fertility VARIABLES (1) (2) Married or in a civil union (ref. cat. “divorced-separated-.widowed”) 0.846*** 0.773*** (0.0631) (0.0639) Never married ref. cat. “divorced-separated-.widowed”) 0.214*** 0.196*** (0.0648) (0.0653) Not religious (ref. cat. “religious”) -0.130*** -0.128*** (0.0281) (0.0281) Gender role index -0.0688*** -0.0819*** (0.0136) (0.0139) Not harmonized in ISCED system (ref. cat. “Lower than upper secondary”) -0.125 -0.133* (0.0762) (0.0765) Secondary Education (ref. cat. “Lower than upper secondary”) 0.0380 0.000561 (0.0701) (0.0704) Higher Education (ref. cat. “Lower than upper secondary”) 0.472*** 0.401*** (0.0714) (0.0719) Belong to an ethnic minority (ref. cat. “yes”) -0.0529 -0.0816 (0.0583) (0.0588) Born in the country (ref.cat. “yes”) 0.159*** 0.174*** (0.0468) (0.0470) Year of the interview (ref. cat. 2004) -0.000560 0.145 (0.0258) (0.117) Temporary employee (ref. cat. “Permanent employee”) -0.202*** (0.0465) Self-employee (ref. cat. “Permanent employee”) 0.0335 (0.0677) Student (ref. cat. “Permanent employee”) -0.945*** (0.0747) Unemployed (ref. cat. “Permanent employee”) -0.0710 (0.0930) Other (ref. cat. “Regular employee”) -0.212*** (0.0687) Temporary employee*year of the interview -0.112* (0.0643) Self-employee*year of the interview 0.0210 (0.0956) Student*year of the interview -0.0759 (0.100) Unemployed*year of the interview -0.308** (0.128) Other*year of the interview 0.00985 (0.0975) Happy 0.0769*** (0.0104) Happy*year of the interview -0.0126 (0.0147) Being childless (ref. cat. “no”) 0.898*** 0.938*** (0.0366) (0.0371) Female (ref. cat “Male”) 0.0576** 0.106*** (0.0265) (0.0273) 25-29 years old (ref. cat. “18-24 years old”) 1.099*** 0.895*** (0.0387) (0.0407) 30-34 years old (ref. cat. “18-24 years old”) 1.037*** 0.810*** (0.0412) (0.0437) 35-40 years old (ref. cat. “18-24 years old”) 0.0977** -0.149*** (0.0441) (0.0469) Variance at country-level .074*** 0.064*** (0.007) (0.007) Observations 21,507 21,393
Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1
The variable of interest in the first model consists in the one representing the year of the interview in
order to detect whether the individuals interviewed in 2010 and the ones in 2004 have different fertility
intentions according to the year. As it can be noticed, the magnitude of the coefficient is close to zero
and, thus, it is not statistically significant. This means that the period of the interview was not significant
in determining the decision of having a child in the following three years.
The other covariates concerned actually background characteristics, which were considered controls and
were not of primary interest. However, here it is a brief summary of the results concerning them. Marital
status presents as reference category the one of the “not married anymore” at the time of the interview
(separated, divorced, widowed). The other two categories refer to individuals being married or in a
legally recognized civil union and to people who are not married, such as cohabiting or single (although,
as underlined in the methodology, there were some problems of computation for those categories). As
expected, those categories have a positive and significant coefficient when they are compared to the
third one. Thus, those who are married and the ones cohabiting or single have higher short-term fertility
intentions than the ones who are not married anymore. Being religious also increases the intention of
having a child in the short-term with respect to not being religious.
The index concerning the gender role has a negative and significant coefficient. This means that having
a more liberal conception of the role of women leads also to a lower intention of having a child within
the timeframe of three years, keeping other variables constant.
The education level of the individual is also an interesting aspect, since it allows dispelling a myth. The
ones with the highest level of education are the only ones who have higher and significant intentions of
having a child in the short-term with respect to those with the lowest level of education. We would not
expect this relationship, since, according to the previous literature, the higher the education of a person,
the lower the willingness of having a child.
The fact of belonging to an ethnic minority does not represent a significant variable to mark a difference
between those who want to have a child in the short-term and those who do not. On the contrary, the
fact of not being born in the country where the interview took place represents a significant
characteristic, which influences positively the intention of having a child in the following three years.
Being a woman rather than a man strengthens significantly the willingness of having a child in the
short- term. The same results are also present when we consider an individual who already has a child
and one who does not. The last category presents higher and significant timing fertility intentions.
According to this model, being more than 25 heightens significantly the intentions of having a child
within a timeframe of three years, although the magnitude of the coefficients of the various considered
age segments decreases as age increases. The most inclined are the ones in the segment between 25 and
29 years old, a little less the ones aged between 30 and 35 and much less the ones between 35 and 40.
The structure of the model, which consists of two error levels, one concerning individuals and one
concerning countries, is justified by the fact that the amount of the variance of the dependent variable
that cannot be explained by the variables already inserted in the model is significantly different from
zero. This means that an individual belonging to a specific country from the selected sample of ESS, on
average, is characterized by a different attitude towards fertility intentions than another one coming
from another country. However, the contribution of the state-level dimension is not particularly high,
since it amounts only around 2.2% of the total variance. The motivation might be twofold (the two
explanations are not mutual exclusive): on the one hand, it could be due to the poor intra-class
correlation among people belonging to the same country (meaning that the differences in fertility
intentions according to the country are not so marked); on the other hand, also the big capability of the
selected variables could explain the phenomenon.
The second models model takes into account the occupational status, well-being and their interaction
terms with the variables of the year. These last two categories of variables are the most interesting for
us, since we are trying to understand the effect of the economic crisis on fertility intentions. The other
variable of interest is the one regarding the year, as before.
This last covariate is still non-significant, although its sign is reversed with respect to the previous
model. The coefficients of the variables representing occupational statuses are computed with respect to
the most protected category of employees, the ones with a permanent contract (an unlimited one) and
comprise the categories of temporary employees, self-employees, students, unemployed and a residual
category, which is not of our interest. The coefficients of temporary employees, students and self-
employees are the expected ones, because the first two are negative and significant, while the third one
is positive but is non-significant. Since the nature of the variable was quite ambiguous, this result is not
surprising.
The findings on the segment of the unemployed are indeed interesting, because the sign of the
coefficient is the expected one, although it is not statistically significant, meaning that the fact of being
unemployed does not imply itself a change in the timing of fertility intentions.
When we look at the results of the interaction terms, they are all of much interest. All the coefficients
have the expected sign, negative for temporary employees, unemployed and students and positive for
the self-employees, but the only significant ones are the unemployed and temporary employees.
However, this last category shows just a weak tie with the possibility of influencing fertility intentions
(it is significant at a statistical level of 10% with a very large sample). The coefficient of well-being
(being more or less happy) is positive and significant, as expected, but its interaction term does not.
For what concerns controls, there are no considerable differences with the previous model, with the
exception of the coefficients regarding the oldest segment of age and the second lowest category of
education. The sign of the first one turns negative and is significant, meaning that being aged between
35 and 40 lowers short-term fertility intentions by a significant amount. This might be due to the
introduction of the terms regarding occupational status, which have been able to capture the age effect
of the most stable jobs.
The coefficient of the second category of education results negative and significant, although its
significance is quite weak (at 10% level in a large sample).
Table 3 shows the same models computed adding the macroeconomic variables, in order to consider the
macroeconomic context.
The model that does not comprise the occupational status, the well-being and the interactions terms
presents some shifts from the other ones. The unemployment rate affects child-timing fertility intentions
in a negative and significant way, while the annual change of GDP also influences the desire of having a
child within a timeframe of three years positively and significantly, although the level of significance is
only at ten percent. There are no significant changes from the similar model without the
macroeconomic variables, with the exception of the reversal of the sign of the coefficient of the year of
the interview, which results significant and positive.
The second model, which consists of macroeconomic variables, occupational status, well-being and
their interactions, fulfill our expectations on the sign and the significance of the coefficients of
unemployment rate and the change of GDP.
The coefficients of the second level of education (according to the ISCED scale) and the ones of the
highest level of age change with respect to the model without the working and well-being variable, as it
occurred in the model presenting only individual-level characteristics.
The most important change regarding the variables concerning occupational statuses regard the fact that
the coefficient of the interaction between the year of the interview and the category of temporary
employees is not significant anymore (although the previous relationship was weak), while the one
concerning unemployed is still statistically significant.
This shows that the crisis impacts on fertility through the employment condition. The crisis amplifies
the effect of the crisis for individuals that are unemployed. That effect (equal to -0.312 in the model (4))
is able for that category to counterbalances the positive sign of the main effect of crisis dummy
(+0.199).
Table 3. Models with both individual and macroeconomic characteristics – Intentions of fertility VARIABLES (3) (4) Unemployment rate -0.0290*** -0.00947*** (0.00336) (0.00336) Percentage change of GDP 0.0121* 0.0239*** (0.00688) (0.00691) Married or in a civil union (ref. cat. “divorced-separated-.widowed”) 0.845*** 0.759*** (0.0631) (0.0639) Never married ref. cat. “divorced-separated-.widowed”) 0.203*** 0.186*** (0.0648) (0.0653) Not religious (ref. cat. “religious”) -0.127*** -0.0517* (0.0275) (0.0271) Gender role index -0.0813*** -0.0755*** (0.0137) (0.0143) Not harmonized in ISCED system (ref. cat. “Lower than upper secondary”) -0.124 -0.139* (0.0763) (0.0767) Secondary Education (ref. cat. “Lower than upper secondary”) 0.0429 0.00315 (0.0702) (0.0705) Higher Education (ref. cat. “Lower than upper secondary”) 0.465*** 0.385*** (0.0716) (0.0721) Belong to an ethnic minority (ref. cat. “yes”) -0.0558 -0.0966* (0.0583) (0.0587) Born in the country (ref.cat. “yes”) 0.137*** 0.161*** (0.0468) (0.0471) Year of the interview (ref. cat. 2004) 0.0613** 0.199* (0.0294) (0.119) Temporary employee (ref. cat. “Permanent employee”) -0.236*** (0.0464) Self-employee (ref. cat. “Permanent employee”) -0.000143 (0.0677) Student (ref. cat. “Permanent employee”) -0.961*** (0.0746) Unemployed (ref. cat. “Permanent employee”) -0.0676 (0.0931) Other (ref. cat. “Permanent employee”) -0.226*** (0.0685) Temporary employee*year of the interview -0.0906 (0.0644) Self-employee*year of the interview 0.0522 (0.0958) Student*year of the interview -0.0510 (0.100) Unemployed*year of the interview -0.312** (0.128) Other *year of the interview 0.0210 (0.0975) Happy 0.0797*** (0.0104) Happy*year of the interview -0.0138 (0.0147) Being childless (ref. cat. “no”) 0.895*** 0.934*** (0.0366) (0.0371) Female (ref. cat “Male”) 0.0575** 0.106*** (0.0265) (0.0273) 25-29 years old (ref. cat. “18-24 years old”) 1.096*** 0.902*** (0.0388) (0.0407) 30-34 years old (ref. cat. “18-24 years old”) 1.026*** 0.813*** (0.0413) (0.0437) 35-40 years old (ref. cat. “18-24 years old”) 0.0872** -0.136*** (0.0443) (0.0471) Variance at country-level .109*** 0.1582*** (.0114) (0.0196) Observations 21,393 21,393 Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1
5. Concluding Remarks
The paper analyzed the impact of Global Crisis on fertility intentions in Europe through a comparison
between ESS data of the second and fifth round that took place respectively in 2004 (four years before
the beginning of financial crisis) and 2010 (after the beginning of the crisis). The choice of those rounds
is given by the presence of a specific and comparable question asking about the intentions of individuals
to give birth to a child within three years from the time of the interview.
The paper used a multilevel logistic regression in order to take into account individual level variables
available in the questionnaires of ESS, as well as macroeconomic indices, in order to evaluate the mixed
effect of micro and macro determinants of fertility plans for European Young Adults. The model
included both, variables related to economic crisis and control variables. In particular, a year dummy
distinguishing observations from the second and fifth rounds has been included in order to evaluate the
difference in fertility intentions across time.
However, the effect of the crisis is supposed to act also through a worsening of macroeconomics
indicators and a general decline in employment rates. Although the time window between the second
and the fifth round of ESS is so long as to allow the presence of external disturbance events, empirical
results are in line with expectations. Even if the crisis dummy is significant and denotes an increase in
fertility intentions in some models, its result has to be read jointly with its interaction with the
employment status: A condition of unemployment is negatively correlated with intentions of giving
birth to a child within three years from the interview, while the interaction for crisis dummy and
unemployment status supports the finding that in crisis times there is an additional negative effect of
unemployment in delay fertility.
Additional results attain the role of macroeconomic indicators. An increase in country level
unemployment rates also has a negative effect on fertility planes.
In conclusion, results are in line with expectations of a depressive effect of global crisis on fertility
expectations.
References
AASSVE, Arnstein; LAPPEGÅRD, Trude. Childcare cash benefits and fertility timing in Norway.
European Journal of Population/Revue européenne de Démographie, 2009, 25.1: 67-88.
ADSERA, Alicia. Changing fertility rates in developed countries. The impact of labor market
institutions. Journal of Population Economics, 2004, 17.1: 17-43.
ADSERA, Alicia. Where are the babies? Labor market conditions and fertility in Europe. European
Journal of Population/Revue européenne de Démographie, 2011, 27.1: 1-32.
AHN, Namkee; MIRA, Pedro. Job bust, baby bust?: Evidence from Spain. Journal of Population
Economics, 2001, 14.3: 505-521.
ANDERSSON, Gunnar. The impact of labour-force participation on childbearing behaviour: pro-
cyclical fertility in Sweden during the 1980s and the 1990s.European Journal of Population/Revue
européenne de démographie, 2000, 16.4: 293-333.
AXINN, William G.; CLARKBERG, Marin E.; THORNTON, Arland. Family influences on family size
preferences. Demography, 1994, 31.1: 65-79.
AXINN, William G.; BARBER, Jennifer S. Mass education and fertility transition.American
Sociological Review, 2001, 66.4: 481-505.
BAIZÁN, Pau; ARNSTEIN, Aassve; BILLARI Francesco C. Cohabitation, marriage, and first birth:
The interrelationship of family formation events in Spain. European Journal of Population/Revue
européenne de Démographie 19.2, 2003: 147-169.
BECKER, Gary S. An economic analysis of fertility. Demographic and economic change in developed
countries. Columbia University Press, 1960: 209-240.
BERNARDI, Laura; KLÄRNER, Andreas; VON DER LIPPE, Holger. Job insecurity and the timing of
parenthood: A comparison between Eastern and Western Germany. European Journal of
Population/Revue Européenne de Démographie, 2008, 24.3: 287-313.
BILLARI, Francesco C., et al. The postponement of childbearing in Europe: driving forces and
implications. Vienna Yearbook of Population Research, 2006: 1-17.
BILLARI, Francesco C.; LIEFBROER, Aart C. Should I stay or should I go? The impact of age norms
on leaving home. Demography, 2007, 44.1: 181-198. 114
BILLARI, Francesco C.; PHILIPOV, Dimiter; TESTA, Maria Rita. Attitudes, norms and perceived
behavioural control: Explaining fertility intentions in Bulgaria. European Journal of Population/Revue
européenne de Démographie, 2009, 25.4: 439-465.
BILLINGSLEY, Sunnee. The post-communist fertility puzzle. Population research and policy review,
2010, 29.2: 193-231.
BUTZ, William P.; WARD, Michael P. The emergence of countercyclical US fertility. The American
Economic Review, 1979, 318-328.
BUTZ, William P.; WARD, Michael P. Will US fertility remain low? A new economic interpretation.
Population and Development Review, 1979, 663-688.
Ciganda, Daniel. Unstable work histories and fertility in France: An adaptation of sequence complexity
measures to employment trajectories. Demographic Research, 2015, 32: 843–876.
DOLADO, Juan J.; GARCÍA‐SERRANO, Carlos; JIMENO, Juan F. Drawing lessons from the boom of
temporary jobs in Spain. The Economic Journal, 2002, 112.480: F270-F295.
EASTERLIN, Richard. Introduction to" Population and Economic Change in Developing Countries".
In: Population and Economic Change in Developing Countries. University of Chicago Press, 1980. p. 1-
4.
EASTERLIN, Richard A. Will raising the incomes of all increase the happiness of all?. Journal of
Economic Behavior & Organization, 1995, 27.1: 35-47.
GOLDSCHEIDER, Calvin. Population, modernization, and social structure, 1971.
GOLDSHEIDER, Frances; GOLDSHEIDER, Calvin. Leaving Home Before Marriage: Ethnicity,
Familism and Generational Relationships Madison, WI: University of Wisconsin Press, 1993.
GOLDSTEIN, Joshua R.; SOBOTKA, Tomáš; JASILIONIENE, Aiva. The End of “Lowest‐Low”
Fertility?. Population and development review, 2009, 35.4: 663-699.
GOLDSTEIN, Joshua R., et al. Fertility reactions to the" Great Recession" in Europe: Recent evidence
from order-specific data. Demographic Research, 2013, 29.
HOEM, Britta. Entry into motherhood in Sweden: the influence of economic factors on the rise and fall
in fertility, 1986-1997. Demographic research, 2000, 2.4: 28.
HOGAN Dennis P.; ASTONE Nan Marie. The Transition to Adulthood. Annual Review of Sociology,
1986, 12: 109-130.
KAHN, Joan R. Immigrant and Native Fertility during the 1980s: Adaptation and Expectations for the
Future. International Migration Review, 1994, 501-519..
KRAVDAL, Øystein. The importance of economic activity, economic potential and economic resources
for the timing of first births in Norway. Population Studies, 1994, 48.2: 249-267.
KREYENFELD, Michaela. Uncertainties in female employment careers and the postponement of
parenthood in Germany. European Sociological Review, 2010, 26.3: 351-366.
KULU, Hill. Migration and fertility: competing hypotheses re-examined.European Journal of
Population/Revue européenne de Démographie, 2005, 21.1: 51-87.
LESTHAEGHE, Ron. The unfolding story of the second demographic transition.Population and
Development Review, 2010, 36.2: 211-251.
LUCI, Angela, and Olivier Thévenon. "Does economic development drive the fertility rebound in
OECD countries?." (2010).
MACUNOVICH, Diane J. The Butz-Ward fertility model in the light of more recent data. Journal of
Human Resources, 1995, 229-255.
MARTIN, Jay. The ultimate vote of confidence’: Fertility rates and economic conditions in Australia,
1976–2000. Australian Social Policy, 2002: 31-54.
MCLANAHAN, Sara. Diverging destinies: How children are faring under the second demographic
transition. Demography, 2004, 41.4: 607-627.
MCQUILLAN, Kevin. When does religion influence fertility?. Population and Development Review,
2004, 30.1: 25-56
MILLER, Warren B.; PASTA, David J. Behavioral Intentions: Which Ones Predict Fertility Behavior in
Married Couples? 1. Journal of Applied Social Psychology, 1995, 25.6: 530-555.
MODENA, Francesca; SABATINI, Fabio. I would if I could: precarious employment and childbearing
intentions in Italy. Review of Economics of the Household, 2012, 10.1: 77-97.
MODENA, Francesca; RONDINELLI Concetta; SABATINI Fabio. Economic Insecurity and Fertility
Planning-the Case of Italy. IARIW-OECD Conference on Economic Insecurity: Paris 2011
Downloaded on March 27th from www. iariw. org/papers/2011/ModenaPaper. pdf, 2011.
ORSAL, Deniz DK; GOLDSTEIN, Joshua R. The increasing importance of economic conditions on
fertility. In: annual meetings of the Population Association of America. Dallas, Texas, April. 2010. p.
15-17.
PERELLI‐HARRIS, Brienna. The Influence of Informal Work and Subjective Well‐Being on
Childbearing in Post‐Soviet Russia. Population and Development Review, 2006, 32.4: 729-753.
RANJAN, Priya. Fertility behaviour under income uncertainty. European Journal of Population/Revue
européenne de Démographie, 1999, 15.1: 25-43.
ROSENWAIKE, Ira. Two generations of Italians in America: their fertility experience. International
Migration Review, 1973, 271-280.
SANTOW, Gigi, et al. Deferment of the first birth and fluctuating fertility in Sweden. European Journal
of Population/Revue européenne de Démographie, 2001, 17.4: 343-363.
SCHMITT, Christian. Gender-specific effects of unemployment on family formation: a cross-national
perspective. Discussion papers//German Institute for Economic Research, 2008.
SOBOTKA, Tomáš; SKIRBEKK, Vegard; PHILIPOV, Dimiter. Economic recession and fertility in the
developed world. Population and Development Review, 2011, 37.2: 267-306.
STEPHEN, Elizabeth Hervey; BEAN, Frank D. Assimilation, disruption and the fertility of Mexican-
origin women in the United States. International Migration Review, 1992, 67-88.
TESTA, Maria Rita; BASTEN, Stuart. Have Lifetime Fertility Intentions Declined During the “Great
Recession”?. 2012.
VAN DE KAA, Dirk J. The idea of a second demographic transition in industrialized countries. Birth,
2002, 35: 45.
VAN DE KAA, Dirk J. Demographic transitions. Demography, Encyclopedia of Life Support Systems
(EOLSS), Developed under the Auspices of the UNESCO (Oxford, UK: Eolss Publishers). http://www.
eolss. net, 2008.
VIKAT, Andres. Women’s labor force attachment and childbearing in Finland.Demographic Research,
2004, 3.8: 175-212.
Recommended